JP3481087B2 - Voice communication fluctuation absorption method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a local area network (hereinafter referred to as "LAN") which is an asynchronous network, an Internet protocol,
In the packet communication network such as the Internet which is a network (hereinafter referred to as “IP”), a delay caused by network traffic or the like (this is referred to as “fluctuation”) particularly when voice communication is performed using the IP network.
The present invention relates to a voice communication fluctuation absorbing method for absorbing noise.

[0002]

2. Description of the Related Art IP networks represented by the Internet are widely used for data communication. Recently, IP
Focusing on economics in network communication, so-called "Internet telephones" that perform voice communication on this IP network are being commercialized. However, in an IP network that has developed as a data communication network, it is difficult to secure voice as much as a normal telephone. Even in the same asynchronous network as this IP network, the asynchronous transfer mode (As
Synchronous Transfer Mode (hereinafter referred to as “ATM”) communication has been considered from the beginning to realize integrated communication of voice and data, but compared to the case where it has been fixed in an IP network developed without being aware of voice communication. There is a problem in securing the voice quality. For example, as a frame unit for data transmission, a frame in a LAN, which is the basis of an IP network, has a fixed-length 53-byte cell in which ATM, which is conscious of voice communication, has a fixed length, whereas medium access in a packet switching system suitable for voice communication is performed. Data on a control (Media Access Control, hereinafter referred to as “MAC”) frame has a variable length of 46 to 1500 bytes, for example.

When the relationship between the IP network and the voice communication is sorted out, the delay becomes large when the communication band secured on the IP network side is narrow. On the contrary, when the communication band secured on the IP network side is wide, the delay is small. Further, when the IP network side receives a momentary burst traffic for file transfer of long information, a momentary delay occurs. On the other hand, in voice communication, the clarity of the call can be secured so that there is no loss of voice.
Also, the less the delay, the more natural conversation can be done. Considering the relationship between the IP network and the voice communication, a voice / data integrated communication system as shown in FIG. 2 has been conventionally proposed. In this integrated voice / data communication system, a voice concentrator (hereinafter referred to as “voice hub”) 4 is connected to a router 2 which is an inter-LAN connecting device provided in the IP network 1 via a transmission line 3, and further, One or a plurality of terminal devices 6 such as general analog telephones and G3 type facsimile devices (hereinafter referred to as “G3 FAX”) are connected to the voice hub 4 via a transmission line 5. The audio hub 4 has a function of interconnecting (gateway) the IP network 1 and the terminal device 6, and has a central processing unit (hereinafter referred to as “CPU”) that controls the entire device, a transmission buffer 4a for accumulating audio data, It is composed of a reception buffer 4b for accumulating voice data, a codec for converting an analog voice signal into a digital signal and transmitting the digital signal, and for reproducing the digital signal into an analog voice signal.
The transmission buffer 4a and the reception buffer 4b are FIFOs (First In First Ou
t) system registers, etc.

From a terminal device 6 such as a general analog telephone, I
When sending a voice signal to a terminal device in the P network 1,
The analog audio signal output from the terminal device 6 is sent to the audio hub 4 via the transmission path 5. In the audio hub 4, the analog audio signal sent from the terminal device 6
awPCM (Palse Code Modulation)
Is encoded to generate audio data expressed in a binary number, converted into a digital signal by a codec, and then stored in the transmission buffer 4a. The CPU in the voice hub 4 reads the voice data stored in the transmission buffer 4a, adds a header that is control information such as a transmission procedure and a destination address to the voice data to generate a voice packet. Is transmitted to the router 2 via the transmission line 3 in the form of a transmittable MAC frame. The MAC frame sent to the router 2 is sent to the destination terminal device in the IP network 1 designated by the header in the voice packet.

FIG. 3 is a waveform diagram for explaining a conventional voice communication fluctuation absorbing method in the system of FIG. With reference to FIG. 3, a method of absorbing fluctuations that occur when a voice signal is transmitted from the IP network 1 to the terminal device 6 will be described. When a voice signal is sent from a terminal device in the IP network 1 to a terminal device 6 in the voice network, a MAC framed voice packet is output from the terminal device in the IP network 1 and the voice packet is a router. 2 is sent to the transmission line 3 in the form of a transmission packet PT. The transmission packet PT sent to the transmission line 3 arrives at the voice hub 4 and arrives at the packet P.
It is received in the form of R. In the audio hub 4, the CPU takes out audio data from the arrival packet PR and sequentially writes it in the reception buffer 4b. Fixed in receive buffer 4b
When the voice data is accumulated up to the voice data amount (that is, the buffer size) BS, the voice data is read from the reception buffer 4b in the written order, reproduced by the codec into an analog voice signal, and transmitted via the transmission line 5. Sent to the terminal device 6. As shown in FIG. 3, when the arrival packet PR includes fluctuations, the fluctuations are absorbed by the write and read operations of the reception buffer 4b. The time difference from the writing to the reading of the reception buffer 4b becomes the reproduction delay time of the analog audio signal reproduced by the codec.

[0006]

However, the conventional voice communication fluctuation absorbing method has the following problems. The transmission buffer 4a and the reception buffer 4b in the voice hub 4 are simple FIs having a short buffer size BS.
It is composed of a FO type buffer, and fluctuations are absorbed by the reception buffer 4b. Arrival packet P
When a delay due to fluctuations occurs in R, the silence state occurs only at the moment when the delay exceeding the past delay time occurs.
The fluctuation absorption buffer size BS follows its maximum delay time. That is, the buffer size BS increases according to the maximum delay time that occurs in the IP network 1. Therefore, it is easy to ensure the clarity of voice. However, if a fluctuation that exceeds the fluctuation absorption time occurs, it cannot be absorbed. Further, since the receiving buffer 4b for absorbing the fluctuation is composed of a simple FIFO type buffer,
Even if there are infrequent burst delays that do not affect the call quality even if there is a loss of voice packets accumulated in it, the delay remains and cannot be recovered until the call is disconnected. Real-time conversation is disturbed. SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for absorbing voice communication fluctuations, which can solve the problems of the above-mentioned conventional techniques, guarantee continuity and clarity of conversation, and perform natural conversation in real time.

[0007]

In order to solve the above-mentioned problems, in the invention according to claim 1 of the present invention, voiced and silent voices are generated from a framed voice packet transmitted from an IP network. Data is extracted and written
FIFO method with constant playback delay time from reading to reading
In the voice communication in which the voice data is sequentially written into the buffer for storing voice data, the voice data is sequentially read with the reproduction delay time delayed in accordance with the written order, and the voice data is reproduced by the decoding means and given to the terminal device. The buffer has a storage capacity large enough to absorb the maximum delay time generated in the IP network , and
A fixed amount of buffer size, which is the amount of accumulated audio data
The function to output the buffer size detection signal is exceeded.
is doing. Then, if the buffer size is
Absorbs the delay time caused by fluctuations
Grows from the buffer to the buffer size
When the detection signal is output, the buffer size
For more than a certain amount for a certain period of time, and
There is a certain amount of silence in the audio data written in the buffer.
When the silent pattern detection unit detects that the pattern
, The buffer size detection signal and the silent pattern detection
The reset signal generated based on the
Reset the buffer to change the buffer size
Clear and reset the silent pattern detector
I am trying. By adopting such a configuration, the following operation is performed. In an IP network, for example, the number of router stages involved in communication, the occurrence frequency of burst data traffic, and the like are unclear.
In such an IP network, the maximum fluctuation time can be from several tens ms to several seconds. Therefore, in the present invention, since the storage capacity of the buffer is set large,
Even if the maximum fluctuation time occurs in the IP network, the operation of the buffer (that is, fluctuation
(The buffer size changes depending on the delay time due to) , and the fluctuation is absorbed. In addition, if a momentary burst delay with a relatively low frequency of occurrence occurs, the reset signal
Resets the buffer and clears the buffer size.
The playback delay time of this buffer is reduced.

According to the invention of claim 2, in the voice communication similar to that of claim 1, the same FIFO system as that of claim 1 is used.
Size of the audio data storage buffer
Is delayed due to fluctuations that occur in voice packets
It absorbs the space and grows,
When the size detection signal is output, the buffer
If the size exceeds a certain amount, it may continue for a certain amount of time.
, And the audio data written in the buffer is
The silent pattern detection unit indicates that the silent pattern will be exceeded for a certain period of time.
If detected, the buffer size detection signal and
A reset generated based on the detection result of the sound pattern detection unit
Reset the buffer to reset the buffer
The size of the silence pattern and reset the silent pattern detector.
I'm trying to set it. By adopting such a configuration, if a delay exceeding the predetermined delay time occurs frequently, if such a delay continues for a certain period of time, the reset signal resets the buffer.
Buffer size is cleared by the
Delay time is reduced.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION First Embodiment FIG. 1 is a schematic configuration diagram showing a voice hub for integrated voice / data communication used in a voice communication fluctuation absorbing method according to a first embodiment of the present invention. The voice / data integrated communication voice hub 40 is provided in place of the voice hub 4 of the voice / data integrated communication system of FIG. 2, and is connected to the IP network 1 via the transmission line 3 and the router 2. , General analog telephone, G3FAX via transmission line 5
Is connected to one or a plurality of terminal devices 6 and the like, and has a gateway function between the IP network 1 and the terminal devices 6 and the like. The audio hub 40 has a CPU 41 for program-controlling the entire apparatus, and a bus 42 for data transmission is connected to the CPU 41. On the bus 42,
The LAN control unit 43 is connected, and the LAN control unit 43 is connected to the router 2 of the IP network 1 via the transmission line 3. The LAN control unit 43 has a function of storing a MAC framed voice packet sent from the router 2, extracting voice data from the voice packet and interrupting the CPU 41, and further transmitting the voice data. It has a function of adding a header to voice data to generate a voice packet, converting the voice packet into a MAC frame, and outputting the MAC frame to the transmission path 3. Here, the audio data is, for example, data encoded by uncompressed ulawPCM from the viewpoint of quality emphasis. The bus 42 includes a FIFO transmission buffer 44 for storing transmission voice data and a FIFO reception buffer 4 for storing reception voice data.
5 is connected. The reception buffer 45 has an extremely large storage capacity capable of absorbing the maximum fluctuation time of several tens ms to several seconds in the IP network 1 in consideration of the fluctuations. These send buffers 44
Also, in the reception buffer 45, the voice data accumulated in the transmission buffer 44 is read by the CPU 41 to
The voice data in the voice packet given to the control unit 43 and stored in the LAN control unit 43 is asynchronously read by the CPU 41 and written in the reception buffer 45. In addition, the reception buffer 45 is stored
The audio data amount (that is, the buffer size) BS is a fixed amount T
When it exceeds H, the buffer size detection signal S45 of "1" is output, and when the reset signal S49 of "1" is input to the reset terminal R, the buffer size BS is cleared. .

An interface unit 47 is connected to the transmission buffer 44 and the reception buffer 45 via a codec 46, and the interface unit 47 is connected to the terminal device 6 via the transmission line 5. The codec 46 is read from the receiving buffer 45 and a code means for converting the analog audio signal sent from the terminal device 6 through the interface unit 47 into audio data of a digital signal by, for example, ulawPCM and giving it to the transmitting buffer 44. The audio data of the digital signal is converted into an analog audio signal by, for example, ulawPCM, and is supplied to the interface unit 47. The interface unit 47 has a function of interfacing between the codec 46 and the terminal device 6. An input terminal of the silence pattern detection unit 48 is connected to the input terminal side of the reception buffer 45, and the output terminal of the silence pattern detection unit 48 and the detection signal S45 output terminal of the reception buffer 45 are connected to a 2-input AND gate 49. Is connected to the input terminal of. The output terminal of the AND gate 49 is connected to the reset terminal R of the reception buffer 45 and the reset terminal R of the silence pattern detection unit 48. The silence pattern detecting section 48 and the AND gate 49 constitute a resetting means.

The silent pattern detector 48 is provided in the reception buffer 4
When the voice data written in the reception buffer 45 has a certain amount of silence pattern, the write data to be written into the reception buffer 45 has a certain amount of silence pattern.
It is a circuit for outputting 48. For example, the reception buffer 45
When the voice data accumulated in the above is encoded by ulawPCM, the silence pattern detection unit 48 monitors whether or not all the lower 7-bit data are “1” (7F (H) or FF (H)). However, it has a function of outputting the silence pattern detection signal S48 of "1" when all become "1", and is configured by a FIFO unit, a silence pattern storage unit, a comparison unit and the like similar to the reception buffer 45. There is. The AND gate 49 outputs the reset signal S49 of "1" when both the input buffer size detection signal S45 and the silence pattern detection signal S48 become "1", and the reception buffer 45 and the silence pattern detection unit 48. Has the function of clearing.

FIG. 4 is a waveform diagram for explaining a voice communication fluctuation absorbing method using the voice hub 40 of FIG. The voice communication fluctuation absorbing method according to the first embodiment will be described with reference to the waveform chart of FIG. When a voice signal is transmitted from the terminal device 6 such as a general analog telephone to the IP network 1, the analog voice signal output from the terminal device 6 is sent to the interface unit 47 in the voice hub 40 via the transmission path 5. . The analog audio signal sent to the interface unit 47 is, for example, ulaw by the codec 46.
It is converted into voice data of a digital signal by PCM and written in the transmission buffer 44. The CPU 41 asynchronously reads the audio data stored in the transmission buffer 44 and sends it to the LAN control unit 43 via the bus 42. LAN
The control unit 43 adds a header to the transmitted voice data to generate a voice packet, and then outputs this voice packet to the MA.
Output in the form of C frame. The output MAC frame is sent to the router 2 via the transmission path 3, and is sent to the other party's terminal device or the like in the IP network 1 based on the header of the voice packet.

Next, a case where a voice signal is transmitted from a certain terminal device in the IP network 1 to the terminal device 6 in the voice network will be described. When a MAC framed voice packet is output from a certain terminal device in the IP network 1, this voice packet is sent to the transmission line 3 via the router 2 in the form of a transmission packet PT. When the transmission packet PT sent to the transmission line 3 arrives at the voice hub 40 in the form of the arrival packet PR, the arrival packet PR is changed to L.
It is stored in the AN control unit 43. When a delay occurs due to traffic of the IP network 1 or the like, this delay is included in the arrival packet PR in the form of fluctuation. CPU4
1 retrieves audio data from the LAN control unit 43 and writes it in the reception buffer 45. The audio data stored in the reception buffer 45 is sequentially read out in the written order and reproduced by the codec 46 into an analog audio signal by, for example, ulawPCM. The reproduced analog audio signal has a reproduction delay time corresponding to the buffer size BS of the reception buffer 45 as shown in FIG. This analog audio signal is sent to the terminal device 6 via the interface unit 47 and the transmission path 5. When there is no voice data in the reception buffer 45 due to a temporary delay (that is, fluctuation) caused by the traffic of the IP network 1 or the like, the CPU 41 causes the reception buffer 4 to receive the voice data.
Since the silence pattern is inserted into the voice signal 5, the analog voice signal sent to the terminal device 6 becomes silent. In order to absorb the delay due to this fluctuation, the buffer size BS of the receiving buffer 45 increases, and the receiving buffer 4
When the audio data stored in 5 exceeds a certain amount TH, the buffer size detection signal S45 output from the reception buffer 45 becomes "1". At this time, the reproduction delay time becomes longer according to the buffer size BS up to the fixed amount TH.

The silence pattern detector 48 is provided in the reception buffer 4
5 monitors the amount of audio data to be written in 5, and the time in which the amount of audio data accumulated in the receiving buffer 45 exceeds a certain amount TH continues for a certain period of time (this time, the buffer 44,
45), and when the sound data written in the reception buffer 45 has a certain amount of silence pattern, the silence pattern detection signal output from the silence pattern detection unit 48. S48 becomes "1". For example, in the silent pattern detection unit 48, if the audio data is encoded by ulawPCM, the lower 7
All bit data is "1" (7F (H) or FF
(H)) is monitored, and when all are "1", the silent pattern detection signal S48 is set to "1". When both the buffer size detection signal S45 and the silence pattern detection signal S48 become "1", the reset signal S49 output from the AND gate 49 becomes "1", and the reception buffer 45 and the silence pattern detection unit 48 are reset. The buffer size BS of the reception buffer 45 is cleared to 0. Therefore, the reproduction delay time of the analog audio signal output from the codec 46 is also shortened.

In the first embodiment, the following (a),
There is an advantage like (b). (A) In the IP network 1, the number of stages of the router 2 involved in communication, the occurrence frequency of burst data traffic, and the like are unclear. Such an IP network 1
In, the maximum fluctuation time can be from several tens of ms to several seconds. Therefore, in the present embodiment, the storage of the reception buffer 45
Since the product capacity is extremely large, even when the maximum fluctuation occurs in the IP network 1, this can be reliably absorbed. Therefore, the continuity and clarity of the conversation can be guaranteed. (B) When a momentary burst delay with a relatively low occurrence frequency that does not affect the call quality even if there is a voice packet drop occurs, the buffer size BS is returned to 0, so the voice delay can be performed early. It recovers and enables real-time conversation.

The silent pattern detecting section 48A is provided in place of the silent pattern detecting section 48 in the audio hub 40 of FIG. 1, and the input side is connected to the input terminal of the receiving pattern 45 of FIG. It has a FIFO unit 48a for voice and a silent pattern storage unit 48b for storing silent patterns. The FIFO unit 48a has, for example, the same configuration as the reception buffer 45, and the comparison unit 48c is connected to the output terminals of the FIFO unit 48a and the silent pattern storage unit 48b. The comparison unit 48c compares the voice data stored in the FIFO unit 48a with the silence pattern stored in the silence pattern storage unit 48b, and when a certain amount of silence pattern stored in the reception buffer 45 is detected, “1” is output. It outputs a silent pattern detection signal S48 of "" and is composed of a gate circuit and the like. This silence pattern detection signal S48 is the same signal as the silence pattern detection signal S48 output from the silence pattern detection unit 48 in FIG. The output unit 48d is connected to the output terminal of the comparison unit 48c. Output unit 48
d receives the silent pattern detection signal S48 and outputs a silent pattern detection signal S48d of "1", for example, when detecting that a certain amount of silent pattern is continuously accumulated in the reception buffer 45 for a certain time TC. It has a function to be given to the AND gate 49 in FIG. 1, and is configured by a register or the like.

FIG. 6 is a waveform diagram for explaining a voice communication fluctuation absorbing method using the silent pattern detecting section 48A of FIG. The voice communication fluctuation absorbing method according to the second embodiment will be described with reference to this waveform diagram. In the first embodiment, when the voice data written in the reception buffer 45 has a certain amount of silence pattern, the silence pattern detection signal S48 output from the silence pattern detection unit 48 (this is output from the comparison unit 48c in FIG. 5). The same signal as the silent pattern detection signal S48 output) becomes "1", the reset signal S49 output from the AND gate 49 becomes "1", and the buffer size BS of the reception buffer 45 is cleared to 0. . In this fluctuation absorbing method, for example, the buffer size BS is returned to 0 when a low-frequency burst delay occurs and exceeds a predetermined delay time. However, a delay exceeding the predetermined delay time is high. In such cases, the clarity of the voice may not be guaranteed and natural conversation may be disturbed. Therefore, in the fluctuation absorbing method according to the second embodiment, the reception buffer 45 is configured by the FIFO unit 48a, the silent pattern storage unit 48b, and the comparison unit 48c.
When a certain amount of silence pattern accumulated in the above is detected, the comparison unit 48c outputs a silence pattern detection signal S48 of "1" and gives it to the output unit 48d. The output unit 48d operates according to the clock signal CK, and when detecting that a certain amount of silent pattern continues for a certain time TC, outputs the silent pattern detection signal S48d of "1", and the AND gate 49 of FIG. Give to. As a result, the reset signal S49 of "1" is output from the AND gate 49, the reception buffer 45 and the silence pattern detection unit 48A are reset, the buffer size BS of the reception buffer 45 is cleared to 0, and the codec 46 outputs it. The reproduction delay of the analog audio signal is reduced. Therefore, when a delay exceeding the predetermined delay time frequently occurs, the intelligibility of the voice is guaranteed and the natural conversation can be continued.

The present invention is not limited to the above embodiment, and various modifications can be made. Examples of this modification include the following (i) and (ii). (I) The audio hub 40 of FIG. 1 may be configured with other circuits. For example, in FIG. 1 or 5, the silence pattern detection units 48 and 48A and the AND gate 49 constitute the reset means, but the reset means is not limited to this. This reset means is used when a write data overflows to the receive buffer 45 for a certain period, when the overflow time exceeds a certain amount, when a silence is detected, or a combination thereof, or packet communication such as the IP network 1 or the like. Various methods such as clearing the buffer size BS of the reception buffer 45 to 0 can be applied by using the result of monitoring the congestion degree of the network as a trigger. (Ii) In the above embodiment, the audio data accumulated in the reception buffer 45 is uncompressed data encoded by, for example, ulawPCM, but the present invention is not limited to ADPCM and CS.
It is also applicable to data encoded by an encoding method such as -ACELP or LD-CELP. Further, the arrival packet PR arriving at the voice hub 40 may have fluctuations even if it is silence-compressed. Therefore, the fluctuation absorption method of the present invention can be applied to such arrival packet PR.

[0019]

As described above in detail, according to the invention of claim 1 of the present invention, the audio of the FIFO system is used.
Since the data storage buffer has a storage capacity large enough to absorb the maximum delay time generated in the IP network, the number of router stages involved in communication, the frequency of occurrence of burst data traffic, etc. are unclear. Even if the maximum fluctuation occurs in a large IP network, this fluctuation can be accurately absorbed. In other words, a natural call is realized by storing not only voiced packets but also silent packets necessary for making a conversation natural in a large capacity buffer. As described above, according to the present invention, the voice packet and the silent packet are accumulated in the buffer having a large capacity with priority given to maintaining the call quality, and only the silent part that does not really affect the call quality is discarded. It is possible to guarantee the continuity and clarity of the. In addition, buffer size
When's exceeds a predetermined amount, silence that the time exceeding the predetermined amount continues for a predetermined time, and that the audio data stored in the buffer becomes constant amount silence pattern
When it is detected by the pattern detector , the reset signal
Since the buffer is reset to clear the buffer size , if there is a momentary burst delay with a relatively low occurrence frequency that does not affect the call quality even if there is a voice packet defect, the playback delay of the buffer The time is adjusted to recover early from voice delays and enable real-time conversation.

According to the invention of claim 2, a FIFO
The audio data storage buffer of this method has a storage capacity large enough to absorb the maximum delay time generated in the IP network.
Since having the same effect as the invention according to claim 1 is obtained. Furthermore, when the buffer size exceeds a certain amount, when the silent pattern detection unit detects that the time exceeding the certain amount continues for a certain period of time and that the accumulated voice data becomes a silent pattern for a certain period of time or more. , Lyce
Since the buffer is reset by the input signal to clear the buffer size , the clarity of the voice is guaranteed even when the delay exceeding the specified delay time occurs frequently. , Able to continue a natural conversation.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a voice hub for voice / data integrated communication used in a first embodiment of the present invention.

FIG. 2 is a schematic configuration diagram of a conventional voice / data integrated communication system.

3 is a waveform diagram for explaining a conventional voice communication fluctuation absorbing method in the system of FIG.

FIG. 4 is a waveform diagram for explaining a voice communication fluctuation absorbing method using the voice hub of FIG.

FIG. 5 is a schematic configuration diagram of a silent pattern detection unit used in the second embodiment of the present invention.

6 is a waveform diagram for explaining a voice communication fluctuation absorption method using the silent pattern detection unit of FIG.

[Explanation of symbols]

1 IP network 2 routers 6 Terminal 40 audio hubs 41 CPU 43 LAN control unit 44 send buffer 45 receive buffer 46 codecs 48, 48A silent pattern detector 48a FIFO section 48b silence pattern storage unit 48c comparison section 48d output section 49 AND gate

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04L 12/56

Claims (2)

(57) [Claims] 1. Voiced and unvoiced voice data is extracted from a framed voice packet sent from an internet protocol network , and written to and read.
First-in-file with a constant playback delay time at
Writes to the buffer for audio data storage of the write- out method sequentially, and according to the written order , delays the playback delay time.
In the voice communication in which the voice data is sequentially read out , reproduced by the decoding means into a voice signal and given to the terminal device, the buffer has a storage capacity large enough to absorb the maximum delay time generated in the internet protocol network.
Amount of audio data that has a capacity and is accumulated in the buffer
If the buffer size exceeds a certain amount, the buffer size
Has a function of outputting a noise detection signal, and the buffer size is generated in the voice packet.
It absorbs the delay time due to
The buffer size detection signal is output from the buffer.
In this case, the buffer size exceeds a certain amount for a certain period of time, and the buffer size is written.
Make sure that a certain amount of rare audio data becomes a silent pattern.
If it detected by the silence pattern detecting unit, the buffer size
Based on the shift detection signal and the detection result of the silent pattern detection unit.
The reset signal generated by the
To clear the buffer size and
A voice communication fluctuation absorbing method characterized by resetting a sound pattern detecting unit . 2. Voiced and unvoiced voice data is extracted from a framed voice packet sent from an internet protocol network and is read from the written data.
First-in-file with a constant playback delay time at
Writes to the buffer for audio data storage of the write- out method sequentially, and according to the written order , delays the playback delay time.
In the voice communication in which the voice data is sequentially read out , reproduced by the decoding means into a voice signal and given to the terminal device, the buffer has a storage capacity large enough to absorb the maximum delay time generated in the internet protocol network.
Amount of audio data that has a capacity and is accumulated in the buffer
If the buffer size exceeds a certain amount, the buffer size
Has a function of outputting a noise detection signal, and the buffer size is generated in the voice packet.
It absorbs the delay time due to
The buffer size detection signal is output from the buffer.
In this case, the buffer size exceeds a certain amount for a certain period of time, and the buffer size is written.
When the silent pattern detection unit detects that the rare audio data becomes a silent pattern for a certain period of time, the buffer
Size detection signal and the detection result of the silent pattern detection unit
The reset signal generated based on
And clear the buffer size
A method for absorbing fluctuations in voice communication, comprising resetting the silent pattern detector .